Dynamic Surface Site Activation: A Rate Limiting Process in Electron

Jul 22, 2013 - Jared Cullen , Alan Bahm , Charlene J. Lobo , Michael J. Ford , and Milos Toth ... Julie A. Spencer , Joseph A. Brannaka , Michael Barc...
0 downloads 0 Views 2MB Size
Research Article www.acsami.org

Dynamic Surface Site Activation: A Rate Limiting Process in Electron Beam Induced Etching Aiden A. Martin, Matthew R. Phillips, and Milos Toth* School of Physics and Advanced Materials, University of Technology, Sydney, P.O. Box 123, Broadway, New South Wales 2007, Australia ABSTRACT: We report a new mechanism that limits the rate of electron beam induced etching (EBIE). Typically, the etch rate is assumed to scale directly with the precursor adsorbate dissociation rate. Here, we show that this is a special case, and that the rate can instead be limited by the concentration of active sites at the surface. Novel etch kinetics are expected if surface sites are activated during EBIE, and observed experimentally using the electron sensitive material ultra nanocrystalline diamond (UNCD). In practice, etch kinetics are of interest because they affect resolution, throughput, proximity effects, and the topography of nanostructures and nanostructured devices fabricated by EBIE.

KEYWORDS: electron beam induced etching, ultra nanocrystalline diamond, reaction kinetics, defect generation, nanofabrication, surface chemistry



INTRODUCTION Gas-mediated electron beam induced etching (EBIE)1−3 is a direct write nanolithography technique used to modify surfaces at nano- and microscales. EBIE proceeds through chemical reactions induced by electron irradiation of a solid substrate exposed to a precursor gas. Surface-adsorbed precursor molecules such as H2O are dissociated by electrons, generating fragments (e.g., O* and OH*)4 that react with a substrate (e.g., C) to produce volatile species (e.g., CO and CO2) that desorb and are removed by a pumping system, thus giving rise to localized chemical dry etching in the vicinity of an electron beam (see Figure 1). Precursors, such as XeF2, Cl2, ClF3, NH3, O2, and H2O, can be used to etch a wide range of materials including graphene, carbon nanotubes, amorphous carbon,5−12 diamond,13−15 and a variety of metals, semiconductors, and insulators.1−3 Nanometer resolution is attainable3 and ∼4 nm has been demonstrated in H2O-mediated EBIE of carbon nanowires on electrically insulating, bulk quartz substrates.5 The technique is analogous to gas-assisted focused ion beam (FIB) milling.3,16 However, EBIE is a chemical process that does not involve sputtering or ion implantation. EBIE resolution and the time-evolution of structures fabricated by EBIE are affected by the electron flux profile at the substrate surface, and by the precursor adsorbate supply and dissociation rates.2,11,17 The flux profile is defined by the diameter and shape of the electron beam, and the spatial distribution of electrons emitted from the substrate. It governs EBIE resolution in the limit of zero depletion (i.e., in the socalled “reaction rate limited” etch regime) where the etch rate scales linearly with electron flux. However, adsorbate depletion makes the etch rate sublinear with electron flux, which in turn © 2013 American Chemical Society

serves to alter (usually decrease) resolution because the etch efficiency decreases with increasing electron flux, which typically decreases with distance away from the electron beam axis. Consequently, much effort has gone into the development of simulators for predictive modeling of EBIE and the related technique of gas-mediated electron beam induced deposition (EBID).1−3,17−22 The models come in a number of varieties, but all are based on assumptions contained in rate equations of the form ∂Na ∂Nα = Λ − k 0Na − + Da∇2 Na ∂t ∂t

(1)

where a and α signify surface-adsorbed precursor molecules (e.g., H2O) and fragments (e.g., O*), respectively, ∂Na/∂t is the rate of change of concentration of precursor adsorbates at each point on the surface, expressed as a sum of fluxes (m−2 s−1) representing adsorption (Λ = sF(1−Θ)), desorption (k0Na), electron induced dissociation (∂N α /∂t) and diffusion (Da∇2Na). N is number density at the surface, F is the gas molecule flux incident onto the substrate, s is the sticking coefficient, Θ is H2O surface coverage (Θ = AaNa, and is typically limited to 1 ML by the Langmuir isotherm), Aa is the area of a single surface site, k0 is the desorption rate, and Da is the diffusion coefficient. The etch rate is given by:

Received: May 30, 2013 Accepted: July 22, 2013 Published: July 22, 2013 8002

dx.doi.org/10.1021/am402083n | ACS Appl. Mater. Interfaces 2013, 5, 8002−8007

ACS Applied Materials & Interfaces

Research Article

during EBIE (model 2). Subsequently, model 2 is adapted to the specific case of site activation caused by electron beam damage of the substrate (model 3), which is shown to be in excellent agreement with EBIE experiments performed using the electron sensitive material ultra nanocrystalline diamond (UNCD).



Figure 1. Simplified schematic of H2O-mediated electron beam induced etching of carbon: (a) H2O adsorption and surface diffusion, (b) generation of O* fragments by incident and emitted electrons, and (c) etching caused by C volatilization by O* adsorbates. Also shown is (d) a 2 μm wide electron beam with a top-hat flux profile and (e) an AFM image of a pit etched in UNCD using a stationary top-hat beam (diameter = 2 μm, depth (zd) = 175 ± 22 nm).

∂Nα = σαfNa ∂t

(2)

∂zd ∂N = Vγ α ∂t ∂t

(3)

METHODS AND MATERIALS

Modeling. Models 1−3 were implemented using numerical methods described elsewhere.3,17 The parameters f and F were measured directly, s was fixed at unity, k0 ≈ 1013 s−1,36 Ea = 0.48 eV, Vγ ≈ 5.70 Å3 (calculated using a density of 3.5 g/cm2), and the area of a H2O molecule37 ≈ 14.8 Å2. All experiments were performed under conditions where adsorbate depletion is negligible. The parameter D was therefore set to zero.3,22 Calculated Na(t) profiles confirmed that the extent of depletion was negligible (